The spring steel strip 50CrVA which is cold rolled was applied to manufacture the diaphragm of the automotive horn by means of sheet metal forming. The combination of the experiments with back-propagation artificial neural network (BPANN) is used to solve the springback problem of the diaphragm. Experiments have shown that a 4-8-1 BPANN is able to predict the springback of the diaphragm successfully, and the network is able to model the relationship between the springback of the diaphragm and the process parameters rationally. BPANN simulation results and experimental ones have shown that the springback of the diaphragm is particularly influenced by such parameters as blank thickness, Young’s modulus, punch radius and yield ratio. Furthermore, the springback of the diaphragm decreases with the increase of blank thickness and Young’s modulus, but increases with the increase of punch radius and yield ratio.